|One artificial intelligence startup company is now helping journalists to make transcriptions in their reports so that they can publish it later on / Photo by: James Selesnick, U.S. Army via Wikimedia Commons|
Artificial intelligence (AI) is slowly becoming a reality for people in many sectors of a country, notable from healthcare to manufacturing. Today, AI opportunities are also being seen as helpful for journalists, particularly when it comes to the pesky process of transcribing.
If you don’t know what transcribing is, it’s a process where a journalist would either type up or write the words of their interviewees for future reference, or if they need to put it on artcards or flash it on news broadcasts.
In a report by Forbes, AI is being utilized to help journalists more efficiently do their transcription work. Traditionally, a transcription is needed before anyone in a news agency can grab hold of important soundbites or quotes. But even if an interview runs for 30 minutes, manual transcription of something like that could take about an hour and a half.
Startup apps like Rev and Scribie aim to help journalists out in this problem and, by extension, also solve the wait time of other people in a news organization dependent on transcriptions, such as social media editors needing Twitter and Facebook content, or even online editors that need to set to work writing news articles with quotes from the interview almost immediately.
Trint, a London-based startup with the exact goal to help journalists out with their transcriptions, has recently come into $4.5 million in April 2019 from investors like the Associated Press, Horizons Lab, and TechNexus. Google is also backing up the startup that excellently “blends a text editor with an audio/video player and then glues the AI-generated text to the source audio to the millisecond, which can also be searched.”
This is not the end-all-be-all solution, though, as CEO Jeff Kofman still says that the app will only likely help journalists and writers secure a more or less 95% to 99% accuracy on the transcription, provided that the audio itself is clean.